Fwd: [ontolog-forum] Watch out Watson: Here comes Amazon Machine Learning - ZDNet - 2015.04.10

[Jack Park]

I think this message from John Sowa has bearing on building Watson-like systems. In fact, I regularly read what he says; it seems clear that many of the things I say about OpenSherlock are reflected in ideas put forth by him.

At the end of Sowa's message is a link to a revised slide deck of his.
That slide deck links to a Minsky paper
http://web.media.mit.edu/~push/CognitiveDiversity.pdf
and mentions a patent
http://www.google.com/patents/US8566321

---------- Forwarded message ----------
From: John F Sowa
Date: Fri, Apr 17, 2015 at 11:28 PM
Subject: Re: [ontolog-forum] Watch out Watson: Here comes Amazon Machine Learning - ZDNet - 2015.04.10
To: ontolo...@ontolog.cim3.net


On 4/17/2015 6:58 AM, Thomas Johnston wrote:
> Is Watson rules-based AI or connectionist  (neural network) AI?

The IBM research project for the Jeopardy! challenge put together
a wide range of technologies for many different tasks that have
to work together.  For any AI paradigm X, if you ask "Is Watson X?",
the answer is either yes, some versions are, or it could be.

In any case, there are many issues involved:

  1. Rule-based systems and connectionist systems have very different
     goals and solve very different problems.  They are complementary
     rather than competing.

  2. Among other things, IBM has also designed a chip with 5.4 billion
     transistors that is specialized for implementing the so-called
     "neural networks" -- which aren't really a realistic model of
     actual neurons.  For a brief summary of the technology, see
http://spectrum.ieee.org/computing/hardware/how-ibm-got-brainlike-efficiency-from-the-truenorth-chip

  3. You can't implement a complete system with today's version of
     neural networks.  Even the IBM chip can only serve as one
     component of a much larger system.  If it were plugged into
     Watson, it could perform many functions quite well, but there
     are many other kinds of AI tasks that would be better handled
     by more conventional software.

I discussed some related issues in the following slides:

    http://www.jfsowa.com/talks/micai.pdf
    Why has AI failed?  And how can it succeed?

John

_________________________________________________________________
Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/
Unsubscribe: mailto:ontolog-f...@ontolog.cim3.net
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J